From 61e5b8949b2f5e8973c58f0dcc7c20c4bdade029 Mon Sep 17 00:00:00 2001 From: tslil Date: Fri, 25 Sep 2026 14:46:31 +0100 Subject: factor out nn_train.py into a separate python project to make it more manageable, add plotting and offline strength fit --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'README.md') diff --git a/README.md b/README.md index 762c366..f20fba2 100644 --- a/README.md +++ b/README.md @@ -22,7 +22,7 @@ Download and extract a recent version of buildroot into the working directory. E Standard adversarial tree search (α-β negamax) with iterative deepening, using a neural network evaluation function for leaves and also a transposition table keyed on the Zobrist hash. -The architecture of the neural network is a standard, two-layer dense network which is trained exclusively through self-play using td-leaf (see `src/nn_train.py`). As input the network is fed the top layer of the board only as well as `flats used/flats remaining` fractions for both players and a single float indicating the parity of the board. +The architecture of the neural network is a standard, two-layer dense network which is trained exclusively through self-play using td-leaf (see the `train/` uv project: `cd train && uv run nn-train`). As input the network is fed the top layer of the board only as well as `flats used/flats remaining` fractions for both players and a single float indicating the parity of the board. ## License -- cgit v1.3.1